The 5’-Terminal Region of the Apocytochrome b Transcript in Crithidia fasciculata Is Successively Edited by Two Guide RNAs in the 3‘ to 5‘ Direction*

We analyzed the chimeric guide RNA (gRNA)-mRNA molecules in Crithidia fasciculata that are predicted to transiently exist in editing of the 5”terminal domain of apocytochrome b (CYb) mRNA, by polymerase chain reaction amplification and DNA sequencing, and ob- tained evidence suggesting that among the 14 editing sites numbered from 3’ to 5’, the sequence in the 3’- half of the sites (3‘ block) was specified by one guide RNA species (gRNA-I) and that in the remaining half of the sites (5’ block) by the other guide RNA species (gRNA-11) and that the direction of editing in each block was 3’ to 5’. The predicted transition site of editing by two gRNAs was between the first and second U residues from the 3’ end within editing site 7. We found that a stretch of the edited sequence in the 3’ block of mRNA could form a stable duplex with a stretch immediately upstream of the guide sequence in gRNA-11. The result leads to a successive editing model that the 3’ block of pre-edited mRNA is first edited by gRNA-I, and after completion of editing, the 5’ portion of gRNA-I1 pairs with the edited mRNA for editing of the 5’ block. The location of of within a

Several species of mitochondrial mRNA in kinetoplastid protozoans such as Crithidia, Leishmania, and Trypanosoma are extensively edited after transcription (Simpson and Shaw, 1989;Benne, 1990;Feagin, 1990;. The location of editing domains, number of editing sites within a single editing domain, and number of U residues to be added or deleted at each editing site are very specific to individual mRNAs. As to the mechanism of RNA editing, a model that small RNAs, named guide RNA or gRNA,' specify the sequence alternation has been proposed, based on the finding that the intergenic regions of maxicircles contain some sequences that can pair with edited mRNA sequences if G:U pairs are allowed . This model was experimentally supported by detection of small RNA species with oligo(U), tails that can hybridize to the corresponding regions of the maxicircles and minicircles  * This work was supported by grants-in-aid for scientific research from the Ministry of Education, Science and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence (s)  ' The abbreviations used are: gRNA, guide RNA; kRNA, kinetoplast RNA; PCR, polymerase chain reaction. Pollard et al., 1990;Sturm and Simpson, 1990b;Pollard and Hajduk, 1991;van der Spek et al., 1991). Identification of chimeric gRNA-mRNA molecules containing U clusters covalently linked at sites of RNA editing further led to a transesterification model (Cech, 1990;Blum et al., 1991;. Nevertheless, the precise mode of action of the gRNA molecules is yet unknown. According to the computer search data, some of a single editing domain is covered by a few different gRNA species Pollard et al., 1990;van der Spek et al., 1991), but its molecular mechanism is also an unsettled question.
The 5"terminal region of the transcript of the apocytochrome b (CYb) cryptogene in Crithidia fasciculata contains 14 editing sites, including the one that generates the AUG initiation codon (Feagin et al., 1988a). These sites are tentatively numbered in the 3' to 5' direction (see Fig. 3). By computer analysis, two candidate gRNAs, named gRNA-I and gRNA-11, have been assigned for editing of sites 1-4 (3' or downstream block) and sites 5-14 (5' or upstream block), respectively (van der Spek et al., 1991). The existence of two gRNA species in cells has also been demonstrated (van der Spek et al., 1991). The editing pattern is well conserved between Leishmania tarentolae and C. fasciculata, except that mRNA in L. tarentolue contains an additional editing site at the most 3' end and that the number of U residues at the most 5' editing site differs (Feagin et al., 1988a). The computer assignments of the editing blocks by gRNA-I and gRNA-I1 are very similar in both strains van der Spek et al., 1991), but no direct evidence supporting these predictions has been presented.
One approach to assessing the role of two gRNAs in editing of a single domain is to examine the reaction intermediates predicted by the transesterification model. Detection of a series of expected intermediates may provide supporting evidence for this mechanism, even though it is not experimentally proved yet. We thus analyzed the kinetoplast RNA (kRNA) of C. fasciculata by PCR amplification and DNA sequencing, assuming that both putative gRNA-I and gRNA-I1 undergo transient covalent interaction with mRNA. As a result, we could identify the chimeric molecules of both gRNA-I and gRNA-I1 that were covalently linked to partially edited mRNA through U clusters. The sequence data suggested that the transition site of mRNA editing from gRNA-I to gRNA-I1 was between the first and second U residues from the 3' end within editing site 7, in disagreement with the computer prediction, and that the direction of editing in each block was 3' to 5'. We noted that the 5' portion of gRNA-I1 could form a stable duplex with a stretch of the mRNA sequence edited by gRNA-I. The result can be inter-887 preted that mRNA editing by gRNA-I1 is initiated only after completion of editing by gRNA-I.

MATERIALS AND METHODS
Preparation of kRNA-Cells of the C. fasciculata strain Cf-CI were cultured at 28 "C with shaking in the Brain Heart Infusion medium (Difco), supplemented with 10 pg/ml hemin and 100 pg/ml streptomycin sulfate. The kinetoplast was purified from late log-phase cells by the Percoll density gradient centrifugation as described (Birkenmeyer and Ray, 1986). The kRNA fraction was isolated from the kinetoplast by SDS lysis, followed by deproteinization with phenol/ chloroform. The ethanol-precipitated pellet was suspended in 10 mM Tris-HC1 (pH 8.0) and 1 mM EDTA, from which contaminated kinetoplast DNA networks were removed by centrifugation for 2 h at 25,000 rpm in the Spinco 50Ti rotor, and kRNA was recovered from the supernatant by ethanol precipitation.
Cloning and DNA Sequencing-The PCR products were cloned into the HincII site of pUC118 (Vieira and Messing, 1987), singlestranded DNA was prepared from white colonies obtained, and the sequences of the inserts were determined using the Sequenase Kit (U. S. Biochemical Corp.) (Sanger et al., 1977;Vieira and Messing, 1987).

Sequence Analysis of Chimeric gRNA-I-mRNA Molecules-
Assuming that gRNA forms a transient covalent linkage with partially edited mRNA, the junction sequences were analyzed by PCR amplification which uses a 3' primer complementary to a downstream region of the pre-edited mRNA and a 5' primer specific to the 5' portion of gRNA-I. Since the 5' end of gRNA-I in C. fasciculata has not been determined, we used the computer-predicted sequence (van der Spek et al., 1990) in the region immediately upstream from the guide sequence (Fig. 1A). The PCR products were cloned in pUC118, and 19 chimeric clones were isolated. When the sequences of individual clones were deduced, 13 different sequence patterns were obtained (Fig. 1A). Each clone contained the gRNA-I sequence, linked to various lengths of oligo(U),, in the 5' portion and the U stretches were further connected to various editing sites of mRNA in the 3' portion. Since the presence of oligo(U), tails in the isolated gRNA molecules has been demonstrated , the chimeric gRNA-mRNA molecules detected can be interpreted as intermediates in activated states during editing process (Blum et al., 1991).
In all the chimeric gRNA-I-mRNA clones, the gRNA sequence was linked to the U clusters at nucleotide position 39 from the 5' end of the gRNA primer (Fig. 1A). The junction site can be regarded as the transcriptional termination site of gRNAs, for the oligo(U), tails are assumed to be nonencoded . The other junctions of the U stretches were distributed between the nucleotides just 3' of editing sites 1 and 7 of partially edited mRNA, although the A, sequence patterns of chimeric gRNA-I-mRNA molecules. PCR products, obtained with a 5' gRNA-I-specific primer and a 3' mRNA-specific primer, are cloned and sequenced, and those in the junction regions are indicated as RNAs, aligned with respect to the fully edited mRNA sequence. The U clusters in the junction are underlined and connected with mRNAs by lines for the alignment purpose. The clones which gave different sequence patterns are numbered as on the left, and occurrence of clones with identical sequences is indicated as frequencies. The DNA sequences coding for the gRNA-I and pre-edited mRNA moieties are shown in the upper and lower parts. The primer sequences used for PCR amplification are in bones, and nucleotide positions in the gRNA moiety were numbered from the 5' end of the primer for convenience. The sequences upstream and downstream from the junction regions in all the clones are identical to those of indicated DNA sequences. The mature edited mRNA sequence as a reference is shown below the junction sequences, on which the number and size of editing sites are indicated and inserted U residues in each site are indicated by lowercase letters. Note that the DNA sequence in the pre-edited mRNA region is indicated below the edited mRNA sequence, so that the sequence has gaps. B, the duplex structure proposed for the sequences of clones 6-13. Solid lines represent standard base pairs and broken lines G:U pairs. Inserted U residues in individual editing sites are shown by lowercase letters. linkage site found in the majority of clones was that just 3' of editing site 7.
Sequence Analysis of Chimeric gRNA-11-mRNA Molecules-The 3' primer used for PCR amplification was the same as that used for analysis of the gRNA-I-mRNA junction. The sequence for the 5' primer was taken from the 5"terminal sequence of gRNA-I1 determined by van der Spek et al. (1991). PCR amplified DNA was cloned, and by analysis of 19 chimeric clones, 17 different sequence patterns were obtained (Fig. 2.4). As in gRNA-I-mRNA chimeras, each clone contained the gRNA-I1 sequences in the 5' portion which were linked to various editing sites of mRNA through U stretches. The junction of gRNA-oligo(U), in the majority of the gRNA-11-mRNA clones (13 out of 19 clones) was found at nucleotide position 53 or 55 from the 5' end of the gRNA-I1 primer. Those in the remaining clones were either truncated or contained unexpected nucleotides in the junction (indicated by dots in Fig. 2 4 ) . Fluctuation of the junction structures observed for gRNA-11-mRNA clones may be due to either ambiguity of transcriptional termination or post-transcriptional modification of gRNA-11. Another possibility to be taken into account would be artifacts in the amplification step. The other junction of oligo(U), in the gRNA-11-mRNA clones were mainly distributed between the nucleotide just 3' of  FIG. 2. A, sequence patterns of chimeric gRNA-11-mRNA molecules. PCR products, obtained with a 5' gRNA-11specific primer and a 3' mRNA-specific primer, are cloned, and those in the junction regions which gave different sequence patterns are indicated as RNAs, aligned with respect to the fully edited mRNA sequence. Unexpected nucleotides found in the gRNA-oligo(U), junctions and the edited mRNA moiety are indicated by dots, and gaps required for sequence alignment are hyphenated. B, the duplex structure proposed for the sequences of clone 12. Other annotations to the figure are identical with those described in the legend to Fig. 1. editing site 7 and the one just 5' of editing site 14, although clones 9 and 13 contained some deletions in the partially edited mRNA moiety, and clones 16 and 17 contained additional nucleotides in the upstream region.
Editing Blocks Directed by gRNA-I and gRNA-II and Direction of Editing-Detection of both gRNA molecules that are covalently linked to partially edited mRNA through oligo(U), strongly supports the transesterification model (Cech, 1990;Blum et al., 1991). Allowing G:U pairing, the gRNA-I sequence in the chimeric molecules can fold back on the edited mRNA sequences from editing site 1 to the first U in editing site 7, as in Fig. lB, and the gRNA-I1 sequence on the remaining mRNA sequence, as in Fig. 2B. According to the transesterification model, the 3'-terminal U of gRNA, that formed a duplex with pre-edited mRNA at the anchor site, first attacks mRNA at the first mismatched base and produces the chimeric molecule by transesterification. The U stretch in the gRNA molecule then pairs with the guide A or G residue of the gRNA itself, and the second transesterification takes place at the next mismatched base. As can be seen in Fig. IA, the oligo(U), in the majority of clones is connected with the G residue just 3' of editing site 7, but the A residue a t position 39 at the gRNA-I-oligo(U), junction can pair with the first U from the 3' end within editing site 7 (Fig. 1B). It is therefore likely that the transition site of editing from gRNA-I to gRNA-I1 is between the first and second U residues from the 3' end within editing site 7. The mRNA moieties that were connected to oligo(U), in all the clones have completely been edited. Thus it is evident that the editing reaction progressively proceeds in the 3' to 5' direction in both the editing blocks. This is essentially consistent with the conclusion deduced from analysis of pre-edited-edited mRNA junctions (Strum and Simpson, 1990a).

DISCUSSION
As has already been discussed by others (Blum et al., 1991, there is a possibility that the identified chimeric gRNA-mRNA molecules are artificial products formed by ligation of the 3' end of gRNA with the 5' end of cleaved mRNA. However, this is unlikely in view ofthe precise location of the junction points along the editing sites. Although there is no direct evidence that the chimeras are true intermediates, our sequence data can be well interpreted on the basis of the transesterification mechanism. When the two gRNA sequences were assigned on the mRNA sequence, we noted that the 12-base-long sequence of edited mRNA, 5'-uGuuAuuuAGAA-3', from editing sites 4-6 can pair with both the guide sequence o f gRNA-I and the 5' moiety of gRNA-I1 (Fig. 3). The gRNA-I1 molecule does not carry any region that can form a stable duplex with pre-edited mRNA. Although the 5' terminus of gRNA-I has not been determined yet, the region upstream from the guide sequence of gRNA-I can pair with the mRNA region just downstream of the 3' block (Fig. 3). Assuming that the regions of gRNA molecules which can form stable duplexes with mRNA provide the anchor sites for mRNA editing (Fig. 3), an editing model emerged is schematically shown in Fig. 4. The editing reaction on mRNA of the CYb cryptogene is first initiated by base pairing-mediated recognition of the pre-edited mRNA sequence with the anchor sequence of gRNA-I. Followed by editing of the 3' block with the guide sequence of gRNA-I, the anchor sequence of gRNA-I1 recognizes the edited mRNA sequence and initiates editing of the 5' block. As the sequences as well as the editing patterns of C. fasciculata and L. tarentolae are very similar (Feagin et al., 1988a;van der Spek et al., 1991), the same mechanism should function in L. tarentolae. Actually, a similar base pairing can be constructed between the edited transcript and the putative gRNA-I1 molecule, if a single base gap is inserted. Although the mechanism involved in switching of binding from gRNA-I to gRNA-I1 is not known, the edited mRNA sequence can form a more stable duplex with the anchor sequence of gRNA-I1 than that with the guide sequence of gRNA-I (Fig. 3).
As to the mechanism of RNA editing of a single editing block, both a consecutive editing  and random editing models (Decker and Sollner-Webb, 1990; have been presented. The former proposed by  is based on the observation that the 5' portion of each gRNA candidate carries a sequence hybridizable to a site just 3' of each editing block. This model, however, does not provide any information about the editing order of the adjacent blocks by different gRNA candidates. Furthermore, as argued by van der Spek et al. (1991), the length of homologous stretches assigned for gRNA-I1 by Blum  FIG. 3. Predicted base pairings between mRNA and gRNAs. The gRNA-I and gRNA-I1 sequences, which were deduced by analysis of chimeric gRNA-mRNAs, are shown above and below the fully edited mRNA sequence with possible base pairings. Standard base pairs are indicated by solid lines and G:U base pairs with broken lines. Number and lines above the sequences represent the 14 editing sites in the CYb mRNA, numbered from 3' to 5'. The AUG codon formed by editing is double-lined. Lowercase u s in the mRNA sequence represent U residues inserted by RNA editing. Predicted anchor regions of gRNAs that can form stable duplexes with mRNA and 3' and 5' editing blocks are indicated above and below the fully edited mRNA sequence.   (1990) is too small to support stable duplexes. As demonstrated in this paper, the 5' portion of gRNA-I1 can form a much more stable duplex with the mRNA sequence specified by gRNA-I. According to our model, editing by gRNA-I1 is initiated only after completion of the gRNA-1directed editing reaction. Sturm and Simpson (1990a) quantitatively analyzed the sequences of unedited-edited junctions of CYb and cytochrome oxidase subunit I11 (COIII) mRNAs in L. tarentolae by PCR amplification which uses the primer set of the 3'edited and 5'-pre-edited sequences of mRNA. Although the intermediates analyzed were quite different from ours, they obtained the junction patterns supporting the consecutive editing process from 3' to 5' in consistency with our observation. With CYb mRNA in L. tarentolae, they identified a strong pausing site of editing in editing site 8, which corresponds to editing site 7 of CYb mRNA in C. fasciculata. This pausing could be due to the transition of editing process from 5 r u i .....

' i U l n 5' gRNA-I1
gRNA-I to gRNA-11. If this is the case, the transition sites in the two strains differ just by one U residue.
In the transcripts of mitochondrial genes for COIII (Feagin et al., 1988b), a subunit of NADH dehydrogenase (Koslowsky et al., 1990), and subunit 6 of ATPase (Bhat et al., 1990) in Trypanosoma brucei, RNA editing is known to occur extensively along the RNA molecules, and each transcript is assumed to be composed of many editing blocks. In fact, three gRNA genes which can guide partially overlapped sequences on the COIII transcript have been identified on the minicircle (Pollard et al., 1990). It is quite likely that the sequential editing mechanism that proposed in this paper by analysis of CYb mRNA is involved in editing of such long editing stretches.